KOMPARASI ALGORITMA PENGENALAN CITRA DALAM MENDETEKSI KUALITAS GABAH BERDASARKAN WARNA DAN BENTUK GABAH

WIBOWO, LEONARDO ARIE (2024) KOMPARASI ALGORITMA PENGENALAN CITRA DALAM MENDETEKSI KUALITAS GABAH BERDASARKAN WARNA DAN BENTUK GABAH. S1 thesis, Universitas Mercu Buana Jakarta.

Preview	Text (HAL COVER) 01 COVER.pdf Download (551kB) | Preview
Preview	Text (ABSTRAK) 02 ABSTRAK.pdf Download (177kB) | Preview
	Text (BAB I) 03 BAB 1.pdf Restricted to Registered users only Download (287kB)
	Text (BAB II) 04 BAB 2.pdf Restricted to Registered users only Download (489kB)
	Text (BAB III) 05 BAB 3.pdf Restricted to Registered users only Download (304kB)
	Text (BAB IV) 06 BAB 4.pdf Restricted to Registered users only Download (504kB)
	Text (BAB V) 07 BAB 5.pdf Restricted to Registered users only Download (103kB)
	Text (DAFTAR PUSTAKA) 08 DAFTAR PUSTAKA.pdf Restricted to Registered users only Download (223kB)
	Text (LAMPIRAN) 09 LAMPIRAN.pdf Restricted to Registered users only Download (935kB)

Abstract

This study compares the performance of four image recognition algorithms—MobileNet, ResNet, EfficientNet, and VGG—for detecting grain quality based on color and shape. The dataset consists of 1931 grain images classified into two classes: “good grain” and “poor grain.” These images were captured using mobile phones under daylight conditions and stored in PNG format. The research methodology includes data collection, pre-processing, model implementation, and data analysis. The dataset is split into training, validation, and test sets. Models are trained using TensorFlow and Keras, with specific hyperparameters tailored to each algorithm. The research results reveal that MobileNet achieves the highest accuracy, reaching 99.48%, with precision, recall, and F1 score all at 99%. ResNet, EfficientNet, and VGG also demonstrate good performance, with accuracies above 98%. In conclusion, MobileNet stands out as the best algorithm for grain quality detection, followed by ResNet, EfficientNet, and VGG. Implementing these algorithms can enhance sorting efficiency and maintain product quality in the rice industry. Kata Kunci: CNN, MobileNet, RestNet, EfficientNet, VGG Penelitian ini membandingkan performa empat algoritma pengenalan citra dalam mendeteksi kualitas gabah berdasarkan warna dan bentuk: MobileNet, ResNet, EfficientNet, dan VGG. Dataset terdiri dari 1931 gambar gabah yang diklasifikasikan menjadi dua kelas: gabah baik dan gabah buruk. Gambar diambil menggunakan telepon seluler pada kondisi siang hari dan disimpan dalam format PNG. Metode penelitian mencakup pengumpulan data, pre-processing, implementasi model, dan analisis data. Dataset dibagi menjadi data pelatihan, validasi, dan pengujian. Model dilatih menggunakan TensorFlow dan Keras dengan parameter khusus untuk masing-masing algoritma. Hasil penelitian menunjukkan MobileNet mencapai akurasi tertinggi sebesar 99,48% dengan precision, recall, dan F1 score sebesar 99%. ResNet, EfficientNet, dan VGG juga menunjukkan performa baik dengan akurasi di atas 98%. Kesimpulan penelitian ini adalah MobileNet merupakan algoritma terbaik untuk mendeteksi kualitas gabah, diikuti oleh ResNet, EfficientNet, dan VGG. Implementasi algoritma ini dapat membantu industri beras meningkatkan efisiensi penyortiran dan menjaga kualitas produk. Kata Kunci: CNN, MobileNet, RestNet, EfficientNet, VGG

Actions (login required)

View Item